BUG: PeriodIndex fails to handle NA, rather than putting NaT in its place (pandas-dev#46673)

Author: kapiliyer

pandas/tests/frame/methods/test_dtypes.py

@@ -83,15 +83,15 @@ def test_dtypes_are_correct_after_column_slice(self):
     @pytest.mark.parametrize(
         "data",
         [
-            DataFrame({"id": [1, 2, 3, 4], "test": [True, pd.NA, pd.NA, False]}),
-            DataFrame({"id": [1, 2, 3, 4], "test": [True, pd.NA, True, False]}),
+            [True, pd.NA, pd.NA, False],
+            [True, pd.NA, True, False],
         ],
     )
     def test_dtypes_are_correct_after_groupby_last(self, data):
         # GH46409
-        data = data.convert_dtypes()
-        result = data.groupby("id").last().test
-        expected = data.set_index("id").test
+        df = DataFrame({"id": [1, 2, 3, 4], "test": data}).convert_dtypes()
+        result = df.groupby("id").last().test
+        expected = df.set_index("id").test
         assert result.dtype == pd.BooleanDtype()
         tm.assert_series_equal(expected, result)